- Bird-users - Mailing Lists

[PATCH 1/2] flex: Avoid REJECT name conflict.
by Joakim Tjernlund 17 May '10

17 May '10

REJECT is a reserved word in flex and REJECT is also used as %token in bison grammar. Rename REJECT to Reject in bison grammar to avoid name conflict. Same thing for ECHO but that does not seem to cause any conflicts ATM so leave as is. Also delete some dead code in flex output by using the apropiate %option statements. Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund(a)transmode.se> --- conf/cf-lex.l | 5 +++-- filter/config.Y | 4 ++-- proto/static/config.Y | 4 ++-- 3 files changed, 7 insertions(+), 6 deletions(-) diff --git a/conf/cf-lex.l b/conf/cf-lex.l index 3fe3c2e..6612d3c 100644 --- a/conf/cf-lex.l +++ b/conf/cf-lex.l @@ -25,7 +25,6 @@ */ %{ -#undef REJECT /* Avoid name clashes */ #include <errno.h> #include <stdlib.h> @@ -71,12 +70,14 @@ linpool *cfg_mem; int (*cf_read_hook)(byte *buf, unsigned int max); #define YY_INPUT(buf,result,max) result = cf_read_hook(buf, max); -#define YY_NO_UNPUT + #define YY_FATAL_ERROR(msg) cf_error(msg) %} %option noyywrap +%option nounput +%option noinput %x COMMENT CCOMM CLI diff --git a/filter/config.Y b/filter/config.Y index 7723658..cab5dbe 100644 --- a/filter/config.Y +++ b/filter/config.Y @@ -28,7 +28,7 @@ static int make_pair(int i1, int i2) CF_DECLS CF_KEYWORDS(FUNCTION, PRINT, PRINTN, UNSET, RETURN, - ACCEPT, REJECT, ERROR, QUITBIRD, + ACCEPT, Reject, ERROR, QUITBIRD, INT, BOOL, IP, PREFIX, PAIR, QUAD, SET, STRING, BGPMASK, BGPPATH, CLIST, IF, THEN, ELSE, CASE, TRUE, FALSE, @@ -506,7 +506,7 @@ term: break_command: QUITBIRD { $$ = F_QUITBIRD; } | ACCEPT { $$ = F_ACCEPT; } - | REJECT { $$ = F_REJECT; } + | Reject { $$ = F_REJECT; } | ERROR { $$ = F_ERROR; } | PRINT { $$ = F_NOP; } | PRINTN { $$ = F_NONL; } diff --git a/proto/static/config.Y b/proto/static/config.Y index a7e5016..20f07ed 100644 --- a/proto/static/config.Y +++ b/proto/static/config.Y @@ -16,7 +16,7 @@ static struct static_route *this_srt; CF_DECLS -CF_KEYWORDS(STATIC, ROUTE, VIA, DROP, REJECT, PROHIBIT, PREFERENCE) +CF_KEYWORDS(STATIC, ROUTE, VIA, DROP, Reject, PROHIBIT, PREFERENCE) CF_GRAMMAR @@ -54,7 +54,7 @@ stat_route: add_tail(&((struct static_config *) this_proto)->iface_routes, &this_srt->n); } | stat_route0 DROP { this_srt->dest = RTD_BLACKHOLE; } - | stat_route0 REJECT { this_srt->dest = RTD_UNREACHABLE; } + | stat_route0 Reject { this_srt->dest = RTD_UNREACHABLE; } | stat_route0 PROHIBIT { this_srt->dest = RTD_PROHIBIT; } ; -- 1.6.4.4

7 37

Send community
by Simone Morandini 14 May '10

14 May '10

Hi all, is it possible to send a specific community from a given RS participant to another RS partecipant? Thanks, Simone. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.

2 1

Malformed AS_PATH
by Simone Morandini 10 May '10

10 May '10

Hi all, I've just started playing around with BIRD: I configured a server with v. 1.2.1 Simple configuration, just to test, with 2 routers: one is ours, the other is a member's router. From what I see, the session between the member and the RS is ok, while the session between our router and the RS cannot be established. The logfile says: 07-05-2010 15:33:43 <RMT> MIX: Received: Malformed AS_PATH: 4002040201315d and birdc says: bird> show protocols all MIX name proto table state since info MIX BGP T_MIX start 15:33 Idle Received: Malformed AS_PATH Description: MIX Preference: 100 Input filter: ACCEPT Output filter: ACCEPT Routes: 0 imported, 0 exported, 0 preferred Route change stats: received rejected filtered ignored accepted Import updates: 0 0 0 0 0 Import withdraws: 0 0 --- 0 0 Export updates: 0 0 0 --- 0 Export withdraws: 0 --- --- --- 0 Pipe protocols are declared in transparent mode. On my peering router the session with the RS goes from Idle to Active, than Idle again. The route server is currently configured with a private AS, but I think it shouldn't be a problem. Thanks in advance, Simone. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.

2 2

MD5 TCP in Linux
by Ondrej Filip 05 May '10

05 May '10

Hi, we made a very interesting observation. For some reason we downgraded Linux kernel do 2.6.24 and MD5 signed BGP session became unstable. BIRD was able to establish session but later it started to send packets with invalid MD5 signature randomly. This caused BGP session drop sometimes. TSO and scatter-gather were switched off. The problem disappeared after upgrade to 2.6.32. So please be careful with Linux kernel versions. Ondrej

1 0

Xtonlsab bug
by Joakim Tjernlund 30 Apr '10

30 Apr '10

Ondrej, this looks buggy: +static inline void htonlsab(void *h, void *n, u16 len) { memcpy(n, h, len); }; +static inline void ntohlsab(void *n, void *h, u16 len) { memcpy(h, n, len); }; memcpy is not defined to handle overlapping memory. Best to add: if (n != h) memcpy(...) Might as well do that for these too: +static inline void htonlsah(struct ospf_lsa_header *h, struct ospf_lsa_header *n) { *n = *h; }; +static inline void ntohlsah(struct ospf_lsa_header *n, struct ospf_lsa_header *h) { *h = *n; };

3 12

CLI ?
by Joakim Tjernlund 28 Apr '10

28 Apr '10

Tried the CLI(birdc) and I am a bit confused as to whait it can do. I only see this when I press ?: bird> ? configure [soft] ["<file>"] Reload configuration debug ... Control protocol debugging via BIRD logs disable <protocol> | "<pattern>" | all Disable protocol down Shut the daemon down dump ... Dump debugging information echo [all | off | <mask>] [<buffer-size>] Configure echoing of log messages enable <protocol> | "<pattern>" | all Enable protocol exit Exit the client help Description of the help system mrtdump ... Control protocol debugging via MRTdump files quit Quit the client reload <protocol> | "<pattern>" | all Reload protocol restart <protocol> | "<pattern>" | all Restart protocol restrict Restrict current CLI session to safe commands show ... Show status information Not a lot, seems that the CLI only can do very limited configuration or am I missing something here. Jocke

7 20

Patch ping
by Joakim Tjernlund 27 Apr '10

27 Apr '10

Haven't seen any action on this patch, forgotten? http://marc.info/?l=bird-users&m=127200916013140&w=2

3 8

[PATCH] flex: Kill warning
by Joakim Tjernlund 27 Apr '10

27 Apr '10

This kills: cf-lex.c:1874: warning: 'yy_fatal_error' defined but not used Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund(a)transmode.se> --- conf/cf-lex.l | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/conf/cf-lex.l b/conf/cf-lex.l index 6612d3c..82f4c52 100644 --- a/conf/cf-lex.l +++ b/conf/cf-lex.l @@ -71,7 +71,7 @@ int (*cf_read_hook)(byte *buf, unsigned int max); #define YY_INPUT(buf,result,max) result = cf_read_hook(buf, max); -#define YY_FATAL_ERROR(msg) cf_error(msg) +#define YY_FATAL_ERROR(msg) do {cf_error(msg); if(0) yy_fatal_error(msg); } while(0) %} -- 1.6.4.4

3 4

fletcher checksum
by Joakim Tjernlund 27 Apr '10

27 Apr '10

I think I mentioned this in passing but the fletcher checksum impl. is fragile. Try changing "u16 length" to "u32 length" and watch it break for no apparent reason. Jocke

3 5

[PATCH 0/5] IP checksum improvements
by Joakim Tjernlund 27 Apr '10

27 Apr '10

Here are a series of performance improvements on the Internet checksum. With these changes applied I get about 20-30% better performance on x86 and PowerPC. Even though we got off on the wrong foot I got curious enough to do some more investigation and I leared more about "add with carry" and how gcc handle them. Feel free to ignore these patches, I just wanted to at least document my findings in patch form. Joakim Tjernlund (5): checksum: improve add32 checksum: Optimize add32() for PowerPC checksum: use pre increment. checksum: optimize loop and get rid of add16() checksum: Optimize first addition. lib/checksum.c | 61 ++++++++++++++++++++++++++++++++----------------------- 1 files changed, 35 insertions(+), 26 deletions(-)

6 19